The long term objectives of this research are to facilitate research into the genetic etiology of cancer by providing core support for (1) developing statistical models to analyze familial dependencies in pedigree data, (2) preparing a package of computer programs based on these models, (3) preparing clear and comprehensive documentation on the use and structure of these programs, and (4) performing simulation experiments to examine the validity, power, efficiency and robustness of these programs to elucidate etiological mechanisms in cancer family data. Segregation and linkage analysis programs will be based on the new regressive models, both univariate and multivariate, which lend themselves to fast computational algorithms. Logistic regression analysis will be extended to be appropriate for the dependencies that occur in family data. Computer programs will be written to be user-friendly, in standard ANSI X3.9 (FORTRAN 77), and carefully documented with the user in mind. Computer simulation will be used to investigate those statistical properties of the methods developed that cannot be determined analytically.